Skip to main content
You can connect datasets to your variables for the following reasons:
  1. Avoid manual copy-pasting of different set of variables values. Rather, hold all of them together and quickly navigate between them to test individual scenarios.
  2. New variables created in the prompt automatically get added as columns in the linked dataset.
  3. Run evaluations on thousands of rows for automated testing.
There is a bi-directional sync between the Editor, Playground and the Linked Dataset. For example, when you create a new variable that it is linked to a dataset in the Editor and define a value in the Variable Editor, Adaline will automatically create a new column in the linked dataset reporting the new variable and its value.

​
Dynamic Columns in Datasets

Datasets can have dynamic columns configured at the column level. You can set a column to be an API variable or prompt variable directly in the dataset, allowing you to:
  • Fetch live data from external HTTP endpoints for specific columns
  • Execute other prompts and use their outputs as column values
  • Run dynamic columns for all rows, failed rows only, or just the first row to test your configuration
When you use a linked dataset in the Playground, dynamic columns are automatically populated with fresh data. Learn more about dynamic columns in datasets.

​
Linking Datasets to Variables

Follow the steps below to link a dataset to your variables:
1

Create a new prompt

Create a prompt in the Editor with at least one variable.Creating prompts with variables
2

Link the dataset to the Playground

Link the dataset to the Playground.Link the dataset to the Playground
3

Run a dataset's row into the Playground

When the dataset is linked to the Playground, you can navigate all the rows. When you select a row, the variables take on the value of the selected row. If you run the prompt in the Playground, the test case run will use the variable values of the selected row.Run a dataset's row into the Playground
4

Verify the columns of the linked dataset

Go inside the linked dataset and verify that the columns match the variables, with respect to the names. With respect to the numbers, a linked dataset can have more columns than the number of variables. The exceeding columns will be ignored in the computation.Verifying a dataset columns